<DocScrutinizer05>
((USB is not realtime, hence, in order to make an USB CNC, you need...)) still doubting that quite a lot
<whitequark>
DocScrutinizer05: well, there's some isochronous shit
<whitequark>
but it's not easy
<whitequark>
wpwrak: okay, read the stuff you wrote
<wpwrak>
you don't want to send real-time commands. just load a set of instructions and process them. if the pc stops unexpectedly, just bring the mill to a halt
<whitequark>
I think I want an adjustable collet
<DocScrutinizer05>
I think wpwrak and me discussed it and I evaluated the maximum speed you can drive it
viric has quit [Ping timeout: 240 seconds]
<DocScrutinizer05>
and the maximum precision you can achieve (due to 'jitter' of USB) which you anyway only want to use in low speed mode
<DocScrutinizer05>
it's always oust one max one step error
<DocScrutinizer05>
just*
viric has joined #qi-hardware
<wpwrak>
you could get away with it but why take chances ? much easier to cat stuff.gcode >/dev/ttyUSB0 and be done with it
<DocScrutinizer05>
and one (micro)step of steper should always be quite below what's max precision of your CNC
<wpwrak>
else you also have to worry about other usb peripherals getting in the way, your process not getting scheduled in time, and all that
<DocScrutinizer05>
errr, doesn't exactly same problem apply for realtime LPT?
<wpwrak>
of course. that's why you need a dedicated pc for it :)
<wpwrak>
if you put it all into the usb critter you can use your laptop
<DocScrutinizer05>
and I still say that's nonsense
<wpwrak>
well, what happens if you miss a deadline ?
<wpwrak>
a general-purpose pc will sometimes do that, especially if your application involves usb.
<DocScrutinizer05>
the max jitter mustn't exceed 50% of step time. IOW you mustn't have >+/-180° phase jitter
<wpwrak>
(without usb you can have a separate real-time subsystem that will be relatively safe. but it can only talk "best effort" to the rest of the system)
<DocScrutinizer05>
max step freq of any such stepper is well below say 5kHz
<wpwrak>
what happens if you get the timing wrong ?
<DocScrutinizer05>
how wrong?
<wpwrak>
let's assume you can detect it and take corrective actions
<DocScrutinizer05>
completely missing a step?
<wpwrak>
wrong = let's say something prevents you from running for 10 ms
<whitequark>
you'd need a linux-rt kernel even for usb one
<DocScrutinizer05>
as long as you sync in with next or 2nd next step to a hard pace raster, nothing will happen. Motor is too slow (aka inert) to do anything bad with one or two wrong steps
wolfspraul has joined #qi-hardware
<whitequark>
unless you offload the movement completely to the controller
<whitequark>
and then it basically needs a gcode interpreter
<DocScrutinizer05>
sorry, that's nonsense
<DocScrutinizer05>
you can run rt processes in any arbitrary standard linux
<wpwrak>
whitequark: i think the gcode interpreter sounds like the way to go. find the subset you actually need.
<whitequark>
why does linux-rt kernel exist then?
<wpwrak>
DocScrutinizer05: but they're not really RT ;-)
<whitequark>
wpwrak: simple subset of gcode can be implemented very quickly
<DocScrutinizer05>
they are sufficiently RT for a low number of processes or even just one process
<whitequark>
a more interesting one which would be compatible with EMC2 is a tad more work, you'd need shit like expressions
<whitequark>
but it is not really hard
<wpwrak>
whitequark: two different things. regular linux "rt" is high scheduling priority. but disabling interrupts or anything that doens't want to give up its time slice will still delay you
<whitequark>
DocScrutinizer05: ^
<DocScrutinizer05>
when you take care to have a driver in kernel to watch for the timing
<wpwrak>
whitequark: the special rt kernels have a special subsystem that has complete priority over the regular kernel
<whitequark>
and EMC2 really runs on the border of possible jitter for my mill
<DocScrutinizer05>
says who?
<whitequark>
EMC2?
<wpwrak>
whitequark: you probably don't need all the bells and whistles from the beginning. and since you have the code, you can just compile it as a checker with which you can dry-run the jobs, see if there's anything you couldn't handle.
<whitequark>
wpwrak: of course
wolfspra1l has quit [Ping timeout: 264 seconds]
<whitequark>
or just upload it to device and let it verify
<whitequark>
I'd want a higher-end F4 anyway for floating point
<whitequark>
that'll make NURBS splines and so very easy
<wpwrak>
whitequark: yup. once you're confident, that's the way to go. have a nice big led to tell you when something it missing. or add a little OLED ;-)
<whitequark>
yeah
<wpwrak>
if you get an F4 then an OLED would be fitting :)
<DocScrutinizer05>
honestly guys, you sound like you know something about software but pretty much zilch about stepper motors
rz2k has quit []
<wpwrak>
DocScrutinizer05: yes, but we know the sw rather well ;-) if a complex io subsystem linux usb is involved, things get messy. now i didn't try to do rt with usb but i did with ide. and while you can get very good results, you won't always win.
<whitequark>
ehehe
<DocScrutinizer05>
of course when you think you can create a 1second metronome like "while true; do beep; sleep 1; done" then you get into trouble with stepper motors and persumed non RT properties of linux
<whitequark>
it's not a problem to run linux-rt today anyway
<whitequark>
debian has it
<DocScrutinizer05>
look, amiga1000 did "realtime stepper motor" all the time, for floppy head positioning
<whitequark>
DocScrutinizer05: well, it essentially had an RTOS then
<whitequark>
linux is way more complex
<DocScrutinizer05>
haha
<DocScrutinizer05>
yeah, you for sure will fail when you think you can control your LPT or USB from a norma userland process directly
<DocScrutinizer05>
no matter what scheduling you use
<DocScrutinizer05>
toldya you need a kernel land module that runs with kernel prio to do the realtime stuff
<wpwrak>
in fact user space is just fine as long as you request a real-time priority
<DocScrutinizer05>
well, my mildew is pretty close to reality then
<whitequark>
The Castle is amazing
<whitequark>
I've read it while doing a certain paperwork-intensive process
<DocScrutinizer05>
and I think pondering about what's "home2 and "freedom" for a bird that never lands is not any weirder than http://de.wikipedia.org/wiki/Der_Bau
<whitequark>
it's like wanting to kill yourself, but now also in your fantasies
arielenter has quit [Quit: Leaving.]
dandon has quit [Ping timeout: 272 seconds]
dandon_ has joined #qi-hardware
dandon_ is now known as dandon
<whitequark>
okay
<whitequark>
magic smoke HAS escaped from the spindle board.
<whitequark>
some guy from forums reported the same thing
<whitequark>
so it's likely not me
<whitequark>
the transformer is HOT
<whitequark>
I can't touch it, it's that hot
<whitequark>
aha, I'm inclined to think that the transformer was breakdown
<whitequark>
the spindle started to spin very quickly on a minimal setting, then there was some cracking noise and a funny smell
<whitequark>
except it's not the smell of exploded semiconductors, it's more the smell of fresh epoxy
<whitequark>
and a fuse is burned
<whitequark>
I'd say electrical breakdown of insulation in the transformer. guess it wasn't rated for 136V after all.
<whitequark>
oh, and there's nothing visibly burned or exploded or just hot on the spindle board.
<whitequark>
ah, no, there's a dead 7812
<whitequark>
INPUT shorted to GND... a very lucky mode of failure. burned the fuse and probably saved the rest of circuits
<DocScrutinizer05>
yes. but it's hard to believe this been caused by a mere 20% "over"voltage
<whitequark>
sure. as I've said, some guy on the forums reported this board exploding, too, I've googled it by the board markings
<whitequark>
although in my case, if the 7812 exploded, it's certainly a big overvoltage
<whitequark>
I guess I can try to turn it on without load and measure the actual AC output of the transformer
<DocScrutinizer05>
7812 is hardened for for input voltages up to 30V iirc, and should shut down on overtemp. And a transformer hardly ever breaks from overvoltage
<whitequark>
let's hope it doesn't burn my house down or something... reminds me I need to get a fire extinguisher
<whitequark>
there's a smoke alarm, but afaik it's not even plugged in, much less working
<whitequark>
russia, duh.
<DocScrutinizer05>
when transformer is of abysmal quality, it might short a turn in primary coil (out of no particular reason), which in turn causes a short and massive temperature increase due to that, which shorts further turns in primary, reducing the number of active turns on primary, thus turning a e.g. 20:1 transformer into a 1:1 transformer
<whitequark>
in this case, will the transformer itself tend to burn down when I use it?
<DocScrutinizer05>
I'd claim warranty repair at seller/manuf, clearly stating that you probed the input voltage with DVM and it been in allowable range for a 110V device
<DocScrutinizer05>
transformers have nothing that burns
<whitequark>
I mean literally, in a fire
<DocScrutinizer05>
they can get pretty hot
<whitequark>
insulation?
<DocScrutinizer05>
but can't burn
<DocScrutinizer05>
(except the very huge ones)
<whitequark>
oh okay. guess I'll just be quick then
<DocScrutinizer05>
well, eventually isolation will evaporate and maybe create flamable fumes
<DocScrutinizer05>
at maybe 300°C
<DocScrutinizer05>
usually *every* transfomer is built in a way so it "fuses" out before anything like this ever happens
<DocScrutinizer05>
also all transformer devices madatory have a real fuse in primary circuit
<whitequark>
it's connected directly to the mains
<DocScrutinizer05>
which should blow as soon as a single turn in transformer shorts
<whitequark>
well, through a switch
<DocScrutinizer05>
sometimes swirch has builtin fuse
<whitequark>
doesn't seem like such a switch
<DocScrutinizer05>
often transformer has builtin fuse
<whitequark>
well, maybe
pcercuei has quit [Quit: dodo]
<whitequark>
actually, hm, zero voltage on output
<DocScrutinizer05>
often chinese build incredibly dangerous crap and still aren't shy to print "CE" on it
<whitequark>
either a fuse or just some of my plugs are shitty (they are)
<whitequark>
well, the *rest* of the device is built very well. good looking PCBs, nonshitty components, good soldering, all wires are accurately wrapped in cambric with the name of circuit printed
<DocScrutinizer05>
I guess the motors are powered by 36V as well?
<whitequark>
a lot of what looks like optocouplers, but can't quickly google the part#
<whitequark>
a 74HC chip, and something that looks like a high-power linear voltage regulator
<wpwrak>
(counterfeit) don't count on it. sharism once got a nice patch of some fairly trivial chips. they were ordered as originals and looked credible enough. yet they were fakes. of course they only found out after making a batch of milkymist with them ...
<whitequark>
no, that board only has one 24VDC input
<DocScrutinizer05>
yes, those definitely are optocouplers
<DocScrutinizer05>
wow
<DocScrutinizer05>
ok, those steppers are not as powerful
<wpwrak>
in the end you may be able to just use a pc power supply ;-)
<whitequark>
yes, I'm getting that feeling
<DocScrutinizer05>
24V since it been there
<DocScrutinizer05>
the motors need it
<whitequark>
mhmm
<whitequark>
lemme check where it gets the motor voltage
<wpwrak>
use -12 V and +12 V ;-)
<whitequark>
wpwrak: PC PSUs have negligible current on -12V
<DocScrutinizer05>
if it was 110V, the'd have used a SPSU with 110V input
<whitequark>
also this mosfet is connected in a way which requires a common ground... it doesn't have an isolated gate
<whitequark>
so no luck with that scheme. but nice try
<DocScrutinizer05>
hmm?
<wpwrak>
well, if you make -12 V your ground and -5 V your "+5 V", you're kinda almost there ;-)
<DocScrutinizer05>
ooh, you mean because of PC PSU ground?
<whitequark>
DocScrutinizer05: well, it's the gate-source voltage which turns it on
<whitequark>
yes
<whitequark>
if the source is at -12V and the gate is at +12V...
<whitequark>
actually, it's out of AMR
<DocScrutinizer05>
don't worry, what's PSU ground doesn't have to be steppercontrol GND
<whitequark>
which is 20V
<whitequark>
mm
<DocScrutinizer05>
you might use 2 pcs 12V PC PSU in series
<whitequark>
DocScrutinizer05: anyway, PC PSUs can supply like 300mA on -12V
<whitequark>
two PSUs would work, yes
<whitequark>
it's kind of a shitty setup though
<whitequark>
and doesn't give 50V too
<DocScrutinizer05>
make sure to power then both up properly before connecting secondary load
<whitequark>
and I don't have two of them
<whitequark>
and I can just use 24V output from my configurable supply
<DocScrutinizer05>
yes, I'd try that
<wpwrak>
what do the 50 V get mangled into ?
<DocScrutinizer05>
should even work for spindle, at low RPM
<whitequark>
so... I think I should check whether the stepper board is dead first
<DocScrutinizer05>
spindle motor
<whitequark>
with configurable PSU
<DocScrutinizer05>
yes
<whitequark>
DocScrutinizer05: (refund) it would be hard to argue that 134V is within operating conditions
<whitequark>
since the block is clearly labeled as 110V-120VAC
<DocScrutinizer05>
I wouldn't argue about that with them
<whitequark>
ah, gotcha
<whitequark>
that transformer is so massive, we're talking for an hour and it's still hot
<whitequark>
such a waste of metal
<DocScrutinizer05>
hehe
<whitequark>
I could probably sell it for good $
<wpwrak>
well, it took it about 2 ? 3 ? hours to overheat ...
<whitequark>
no, I played with it for about 20 minutes this session
<DocScrutinizer05>
when it actually *over*heated then you'd see that from outside
<whitequark>
and a few hours before that, but it got time to cool down
<wpwrak>
ah. so you russians got a bit more careful after chernobyl :)
<DocScrutinizer05>
and that particular usage pattern makes it *very* unlikely it overheated in 20min, after working 3h before
<whitequark>
hahaha
<whitequark>
never was this joke more appropriate
<wpwrak>
DocScrutinizer05: okay ... became the dominant heating element in whitequark's vicinity :)
<whitequark>
ooooh, THAT's why the room was so hot
<whitequark>
that actually may make perfect sense
<DocScrutinizer05>
[2014-01-25 20:53:43] <DocScrutinizer05> occasional manual control of temperature of whole thing can't hurt, until you got familiar with it
<wpwrak>
;-)
<whitequark>
(I'm not serious about the room, it was just the A/C heating it up. But it might have contribued to the problem)
<whitequark>
(manual control) I did control it, without disassembling the block. the metallic case was a bit warm.
<whitequark>
since metal is a good heat conductor I decided the insides aren't particularly hotter
<DocScrutinizer05>
then it's highly unlikely it overheated
<DocScrutinizer05>
you usually also smell overheated trafo an hour before it becomes critical
<whitequark>
the metal transformer core is screwed to a metal pad which is screwed directly to the metal case with high contact area
<whitequark>
ah, no, there was definitely no smell, even seconds before breakdown
<whitequark>
and I smelled it very strongly immediately after
<whitequark>
so... what was it if not overheat?
<DocScrutinizer05>
crappy isolation of winding wire
<DocScrutinizer05>
the isolation should withstand the 5 times overvolatge at least before breakdown
<wpwrak>
volatile components considered unnecessary unless proven otherwise ?
<whitequark>
wpwrak: volatile?
<wpwrak>
stuff that escapes. under certain acceptable (unless proven otherwise) operating conditions :)
<DocScrutinizer05>
when they used cheap wire to do the transformer coils, and that cheap wire then gets mechanical damage during sloppy manufacturing, then the isolation may break while manufacturing and finally breaks down shorting one wind
<DocScrutinizer05>
or a few winds
<DocScrutinizer05>
s/wind/turn/
<qi-bot>
DocScrutinizer05 meant: "or a few turns"
<whitequark>
right
<whitequark>
btw I can't get off the clamp of the transformer input wire off the terminal
<whitequark>
it's the same type of terminal like e.g. 12V terminals in cars
<DocScrutinizer05>
often the contact wires are soldered to the thinner internal coil wide, and often those solder joints are sharp
<whitequark>
do they have some hidden latches or something? no amount of forces disconnects it
<whitequark>
except it's on the power socket, like mains socket on a computer PSU
<DocScrutinizer05>
either a bumb that fits into hole, or even a real latch
<DocScrutinizer05>
usually bump though, that goes into the hole
<whitequark>
ah, I see it now. a bump.
<DocScrutinizer05>
apply force
<whitequark>
already applying it, with a makeshift lever out of hex key
<whitequark>
seems like the connector would break first
<whitequark>
aha, got it off
<whitequark>
needed more force
<whitequark>
hmm, I could fit a small computer PSU in place of that enormous chunk of metal...
<whitequark>
maybe two
<DocScrutinizer05>
pro tip: the other side of the hole in the male end is usually open
<DocScrutinizer05>
a smal nail...
<whitequark>
... in the head of whoever made that transformer...
<DocScrutinizer05>
hehe
<DocScrutinizer05>
also those plugs always are with two almost closed "tubes", each long side one. You can insert a nail into the "tube" and slightly bend it open. Plug falls off like a dead fly from the wall
<whitequark>
yep, that was my second option
<whitequark>
ok, in steady state it draws 4mA@24V and the LED on the board is glowing
<DocScrutinizer05>
ok, I got a shiny new camera. Let's see how it performs
<whitequark>
wait, why exactly is there a LED in a completely black case?
<DocScrutinizer05>
haha
<DocScrutinizer05>
just for you ;-)
<wpwrak>
it's the flashlight. now look for the camera.
panda|x201 has joined #qi-hardware
<whitequark>
ok, with steppers connected the board tries to draw more than 2.5A of current without connected LPT
<wpwrak>
first was with the camera sitting on a my monitor and with delayed shutter release. for the second i held the camera but used the monitor as support. exposure 1 second, a bit too long for my unsteady hands.
<wpwrak>
hmm. for night mode, i would have to get out a tripod. too much wind :-(
<DocScrutinizer05>
that other thing realy is a 10R?
<DocScrutinizer05>
maybe nit's a Zener then
<whitequark>
color coding on other thing doesn't make sense
<whitequark>
brown black black silver brown
<whitequark>
it's either 1Ω1% or 180Ω1%
<whitequark>
depending on the order
<DocScrutinizer05>
you're talking about the thing at ne555 output?
<whitequark>
yes
<whitequark>
blue body cylindrical resistor
<DocScrutinizer05>
according to layout/componentplacement it's 10R
<whitequark>
yeah, guess that is incorrect
<whitequark>
lemme measure it, actually. mosfet is as good as floating
<DocScrutinizer05>
brown black black, solver
<DocScrutinizer05>
silver
<whitequark>
10R
<whitequark>
indeed
<whitequark>
but the color coding doesn't match
<DocScrutinizer05>
nuke the right brown
<DocScrutinizer05>
10%
<whitequark>
right
<whitequark>
something I'm not aware of? weird encoding scheme?
<DocScrutinizer05>
nah, maybe 1W
<whitequark>
ah.
<whitequark>
the body looks like 1/4W
<DocScrutinizer05>
this stuff is not +really* standardized
<DocScrutinizer05>
could even be a coil/choke :-o
<whitequark>
hahaha
<whitequark>
no, it's marked R2
<DocScrutinizer05>
mhm
<DocScrutinizer05>
what the heck is this huge blob next to the powerswitch?
<DocScrutinizer05>
(K1)
<DocScrutinizer05>
connector?
<whitequark>
right down from K1? yes, it's for the steppers daughterboard
<DocScrutinizer05>
:nod:
<DocScrutinizer05>
I still wonder how that shite works
<DocScrutinizer05>
what's the PWM in/output for?
<whitequark>
unconnected
<DocScrutinizer05>
mhm
<DocScrutinizer05>
o.O
<whitequark>
you can set SW1 to left/right with a jumper
<whitequark>
like on PC motherboards
<whitequark>
it's set to "down", next to Pr1
<whitequark>
I suppose this selects the potentiometer controlled mod
<whitequark>
*mode
<DocScrutinizer05>
ooh I see
<whitequark>
if you set it to "up" it would I guess get the PWM
<DocScrutinizer05>
yes
<DocScrutinizer05>
via optocoupler
<whitequark>
and on the stepper board there's actually an unpopulated PWM output
<whitequark>
DocScrutinizer05: I'm horrified to think what happens if you assert STOP
<whitequark>
my bet: EL817 explodes
<whitequark>
it's rated at 60mA cont, 1A peak
<DocScrutinizer05>
haha
<DocScrutinizer05>
a relais?
<whitequark>
relais?
<whitequark>
oh, relay
<DocScrutinizer05>
I think EL817 must also be some sort of optocoupler or relais
<whitequark>
no. it's an optocoupler. BC817 with a LED pointing at base
<DocScrutinizer05>
mhm
<DocScrutinizer05>
it looks like it operates on spindle running or overloeded
<DocScrutinizer05>
R3/R4 = 1R?
<whitequark>
yes
<DocScrutinizer05>
then it will prolly start on 4A
<whitequark>
no no
<whitequark>
hm
<whitequark>
ah you're right
<DocScrutinizer05>
so it signals overload
<whitequark>
I thought it was plugged in reverse direction
<whitequark>
that makes sense
<whitequark>
also not connected to anything :]
<DocScrutinizer05>
sure, when PWN NC then STOP NC too
<whitequark>
right
<whitequark>
okay. can you explain how the contraption around Rp1 works?
<whitequark>
I have absolutely no idea
<DocScrutinizer05>
err yeah
<DocScrutinizer05>
that's the R that defines how fast the C of NE55 charges and discharges. Funny enough the split charge and discharge on both sides of same poti
<DocScrutinizer05>
so the longer it's "on" the shorter it's "off"
<DocScrutinizer05>
D1 and D2 must be "antiparallel"
<whitequark>
they are
<DocScrutinizer05>
C is charging via D1 and discharging via D2
<whitequark>
ooooo, I understand
<whitequark>
so it's basically changing duty cycle with two diodes, one cap, one pot and one ne555
<whitequark>
that's cool
<whitequark>
without changing freq that is
<whitequark>
really clever
<DocScrutinizer05>
nifty
<DocScrutinizer05>
and PWM abuses NE555 as a schmitt trigger basically
<whitequark>
yeah
<DocScrutinizer05>
I'm lost WTF D3 is doing
<DocScrutinizer05>
oooh
<DocScrutinizer05>
flyback diode for 7812
<whitequark>
hm
<DocScrutinizer05>
ok, eaten it
<DocScrutinizer05>
C8 is the freq
jekhor has joined #qi-hardware
<whitequark>
mhm, can't seem to easily find dual 24V/48V power supply in .RU
<whitequark>
certainly not at these currents
<whitequark>
DocScrutinizer05: how hard it is, would you think, to make a mains 24V/48V switchmode supply myself?
<whitequark>
I don't really want another 3kg shitty transformer there which is likely to burn just as well!
<DocScrutinizer05>
I recommend to NOT try that
<whitequark>
you think so?
<DocScrutinizer05>
building SPSU is no witchcraft but annoyingly tricky
<whitequark>
well, I've built several. but for lower voltages.
<DocScrutinizer05>
starts with transformer and doesn't end at low-ESR Cs
<whitequark>
okay, bad idea, I agree
<whitequark>
how about this: make a 48V->24V buck converter and use a single 48V off the shelf supply?
<whitequark>
or just buy it
<DocScrutinizer05>
how about using two 24V off-the-shelf supplies
<whitequark>
too much shit to bother about
<whitequark>
really don't like this as a permanent solution
<DocScrutinizer05>
what's the problem?
<DocScrutinizer05>
24V+24V=48V
<whitequark>
need to carry two supplies, remember to plug them both in before enabling the device
<DocScrutinizer05>
nah, they fit in where the transformer been
<whitequark>
two? unlikely
<whitequark>
well, maybe
<whitequark>
do you have a specific PSU in mind?
<DocScrutinizer05>
laptop PSU?
<whitequark>
won't fit for sure. also, expensive
<whitequark>
also, not enough current, those I've seen were under 100W
<whitequark>
and spindle alone needs 200W... steppers eat about 50W
<DocScrutinizer05>
Iseem to have seen them up to 3.5A
<whitequark>
with what voltage?
<DocScrutinizer05>
well, I have no particular PSU in mind. not really
<whitequark>
let's say I don't know where or which PSU to get to satisfy my constraints
<whitequark>
never seen anything that'd fit
<whitequark>
also, 250W @ 48V = 5.2A
<DocScrutinizer05>
alibaba probably right place to search for that kind of stuff
<whitequark>
"Functionality/Degree of Testing: Unit powers on and indicates “EMG”. We do not have the necessary resources to further test this unit. Power cable is not included."
<whitequark>
okay. let's first wait and see what the CNC vendor responds
<whitequark>
if they just send me a new control block, awesome. otherwise I'm telling that this one burned
wej has quit [Ping timeout: 245 seconds]
wej has joined #qi-hardware
wolfspra1l has quit [Quit: leaving]
xiangfu has joined #qi-hardware
wolfspraul has joined #qi-hardware
xiangfu has quit [Ping timeout: 260 seconds]
xiangfu has joined #qi-hardware
jekhor has quit [Ping timeout: 252 seconds]
FDCX has joined #qi-hardware
xiangfu has quit [Ping timeout: 265 seconds]
xiangfu has joined #qi-hardware
bartbes has quit [Quit: No Ping reply in 180 seconds.]
bartbes has joined #qi-hardware
<wpwrak>
whitequark: you mentioned that someone else had the same problem. did he mention whether/how he solved it ?
dos1 has joined #qi-hardware
<apelete>
hello there
<apelete>
I'm looking for a new soldering iron to replace a cheap 30W model I bought off ebay
<apelete>
something with a fine tip, to be used for electronic parts, I'm thinking about a 60W model maybe
<apelete>
any advice ?
<wpwrak>
apelete: you should consider one with adjustable power. there are often situations where that is useful. that way, you can solder ic pins with low power, reducing risk of damage, and turn up the heat when doing things that have a lot of metal, e.g., connectors or ground planes
<wpwrak>
they cover the medium to high range in terms of price. so you won't find the ultra-cheap stuff that may or may not work. and they also have things with very scary prices :)
<apelete>
wpwrak: do they ship to worldwide ?
<apelete>
it looks great, I hope they do. I might go with the cheap LF-389D model
jekhor has joined #qi-hardware
<wpwrak>
you should be able to find some of these products also in europe
<wpwrak>
they do ship worldwide. i bought my 988D from them many years ago. wasn't even too expensive (these things can be heavy, especially the bigger ones)
<apelete>
wpwrak: problem is, the input power in europe is 220-240V
jekhor has quit [Ping timeout: 245 seconds]
jekhor has joined #qi-hardware
rz2k has joined #qi-hardware
pcercuei has joined #qi-hardware
<CYB3R>
pcercuei: I wanted to ask about some constants in UBIBoot. My RAM chip bus width is 16bit, it has 4 banks, 13 bits for rows, 10 bits for cols.
<CYB3R>
I'll set it like this: SDRAM_BW16 1; SDRAM_BANK4 1; SDRAM_ROW 13; SDRAM_COL 10.
<CYB3R>
But there is also SDRAM_BANK40, SDRAM_ROW0, SDRAM_COL0. What should it be?
<pcercuei>
I'm not sure, that code is not mine
<pcercuei>
but it looks like reset values
<pcercuei>
in board-a320.c, dmcr0 is calculated from SDRAM_BANK40, SDRAM_ROW0, SDRAM_COL0
<pcercuei>
and REG_EMC_DMCR is first set to dmcr0, then to dmcr
<pcercuei>
I'm don't know why it does that but it's probably needed
<pcercuei>
so SDRAM_ROW0=11 and SDRAM_COL0=8 would be the minimal row/column number the SoC can handle
<pcercuei>
which would make sense
<CYB3R>
Looks like SDRAM_SMTH0 is SDRAM_SMTH - 1
<pcercuei>
so just don't change those
<pcercuei>
no, I think it's just a coincidence
<nicksydney>
wpwrak: nice view from that camera
<CYB3R>
Okay, I'll leave those values as they are.
<CYB3R>
SDRAM_TRWL should be the value of "WRITE recovery time" from my datasheet
<wpwrak>
apelete: they tend to have 220 V options
<apelete>
wpwrak: you mean the stations actually come with 110 to 220V power regulators built in ?
<apelete>
it seems to be the same one but with 220V instead of 110V
<apelete>
I asked if it comes with a EU plug, will buy one depending on the reply
<wpwrak>
apelete: looks good
<wpwrak>
(110 vs. 220 V) no, they're usually for one voltage only. but howard often have (or at least had) a 220 V version, too.
<apelete>
okay
<apelete>
wpwrak: btw, what's the benefit of having a higher power iron (eg. 60W vs 45W or 30W) when one can set the temperature like on the Xytronic ?
<CYB3R>
pcercuei: and what value should SDRAM_TREF be? My datasheet says: "Refresh period (8,192 rows) (tREF) max = 64 ms"
porchao has quit [Quit: Leaving...]
porchao has joined #qi-hardware
pcercuei_ has joined #qi-hardware
pcercuei has quit [Ping timeout: 240 seconds]
dos1 has quit [Quit: Kabum!]
dos1 has joined #qi-hardware
xiangfu has quit [Remote host closed the connection]
pcercuei_ has quit [Read error: Operation timed out]
rz2k has quit [Read error: No route to host]
jekhor has quit [Ping timeout: 240 seconds]
CYB3R has quit [Ping timeout: 245 seconds]
pcercuei has joined #qi-hardware
rz2k has joined #qi-hardware
<wpwrak>
apelete: more power = can heat up quicker, can recover from a heat loss more rapidly, can maintain higher temperatures, can heat larger areas
<wpwrak>
apelete: the main real-life benefit is the ability to heat larger areas. e.g., if you need to solder things on a large ground plane that ground plane will suck up a lot of heat. but with 60 W you'll already have some flexibility
<wpwrak>
e.g., mine has 60 W and i rarely have troubles with it getting too cold. besides, reasonable layouts avoid situations where you'd be forced to solder directly on a ground plane
<whitequark>
wpwrak: (similar problem) he's got a resistor and 555 blown out. I assumed the underlying problem was the same due to similar symptoms
<whitequark>
and no, the forum thread just dies
<whitequark>
*died
<wpwrak>
hmm, pity. oh, and responses from .cn may take while due to CNY.
<apelete>
wpwrak: wow, never thought about those technical details about soldering, quite a knowledge you've got there
<wpwrak>
apelete: it's just what you notice as you solder a few things ;-)
<apelete>
wpwrak: well, let's hope having a better tools will help me get better results, I was terrible last time on the nanonote
<whitequark>
wpwrak: ah, that's why the vendor doesn't respond... yeeeah
<apelete>
s/tools/tool/
<qi-bot>
apelete meant: "wpwrak: well, let's hope having a better tool will help me get better results, I was terrible last time on the nanonote"
<whitequark>
when exactly it is this year?
<wpwrak>
yeha, bad tools can make anything a nightmare. and often it's just so bad that you can't even get a clear picture of why you're having a problem
<wpwrak>
whitequark: this week :) the new year started on the 24th, i think
<whitequark>
ok, then I will not start a ebay dispute when it suggests (at 28th)... give them the benefit of holidays
<whitequark>
it's one week right?
<wpwrak>
yup
Luke-Jr has quit [Excess Flood]
Luke-Jr has joined #qi-hardware
porchaso0 has joined #qi-hardware
porchao has quit [Ping timeout: 260 seconds]
<whitequark>
wpwrak: what CAD do you use? librecad?
porchao has joined #qi-hardware
porchaso0 has quit [Ping timeout: 260 seconds]
jekhor has joined #qi-hardware
<wpwrak>
whitequark: fped, and in very hackish ways :)
<wpwrak>
oh, for 2D. definitely fped.
<wpwrak>
it lacks some common 2D CAD operations but being parametric makes up for that
<wpwrak>
for 3D, things get messy, though. i plan to have a look at FreeCAD again for that. fped can do 3D (with a little help), but it's rather non-intuitive
<wpwrak>
what you basically do is that you draw slices of the model. and then you give each a certain height. so it's kinda 2.5D. also, since you'll repeat some primitives a lot of times, the gui isn't very useful in this mode of operation.
<whitequark>
mm
<wpwrak>
the killer feature is of course that it's parametric and that this is a "natural" part of the system
<whitequark>
isn't it, um, slightly unfinished?
<wpwrak>
other cad systems often have several interfaces for the same task. then you can choose between "draw", "equation solver", etc.
<wpwrak>
well, is anything ever finished ? :)
<whitequark>
last time I tried to use it, I found it rather incomprehensible
<whitequark>
maybe should make another try
<wpwrak>
what did you expect to find but didn't ?
<wpwrak>
some limitations as CAD come from it being meant to design footprints. that's why i didn't include overly fancy geometric operations.
<whitequark>
that was an overall feeling. well, I guess I'll give it another try
<eintopf>
hey guys
_whitelogger has joined #qi-hardware
whitequark has joined #qi-hardware
wej has joined #qi-hardware
<whitequark>
DocScrutinizer05: actually, there IS a foss way to generate highly efficient toolpaths!
<whitequark>
yep! I'm currently evaluating various foss CADs/CAMs
<whitequark>
so far heekscad/heekscam looks the nicest
wolfspraul has joined #qi-hardware
<DocScrutinizer05>
I find it a tad difficult to understand what exactly been the parameters that resulted in a picture like the one on the URL you posted, but hey...
<DocScrutinizer05>
you can guess what's the tool shape
<DocScrutinizer05>
and what's the rationale for the path, resulting from that
<paul_boddie>
So I started using OpenSCAD more seriously instead.
<whitequark>
so far I'm liking it a lot, it seems both powerful and quite understandable
<whitequark>
and I don't exactly like openscad's scripting approach
<whitequark>
[for me] textual programs work for abstract stuff like networks and algorithms, but not physical objects
<paul_boddie>
FreeCAD has some nice features, albeit confusingly presented.
<paul_boddie>
One thing that shows the difference between the two is the thing known as filleting where you smooth off edges.
<paul_boddie>
In FreeCAD there are two tools for that - you're not supposed to use "the wrong one", which I managed to avoid by chance - whereas in OpenSCAD it's an FAQ.
<paul_boddie>
I figured it out in OpenSCAD, anyway: you just cut a cylinder out of a block and then substract the remainder from whatever you want to fillet.
<paul_boddie>
But yes, there's a lot of sitting and thinking (and using your fingers to visualise things, "old school") with OpenSCAD.
<paul_boddie>
And I'm not sure how well suited it is to producing models that can be machined using traditional tools, although wpwrak may have something to say about that.
<wpwrak>
the fillet ting in openscad means of course that you can't just "fillet" an edge but that you have to consider the whole geometry. this can get messy.
<wpwrak>
paul_boddie: haven't tried to real-life machine from either. only a bit from heekscad and then a lot with fped
<paul_boddie>
I was doing a design the other day, and I was generally thinking in terms of how you'd machine it.
<paul_boddie>
But I'm not sure if there are tools to try and convert the operations to actual machining operations.
<paul_boddie>
I think you'd have to declare such things explicitly, anyway, since CSG allows for some pretty strange stuff.
<paul_boddie>
As in that filleting where the machine would happily cut out the cube, but then flip out upon being told to "negate" it to round off edges. :-)
<paul_boddie>
Specify an antimatter cube to begin with...
<wpwrak>
ah no, the cam part of the cad system should translate that ;-)
<wpwrak>
antimatter cube is a possibility. but you need pretty strong containment fields ...
<paul_boddie>
Easier to go the 3D printer route at that point. ;-)
<whitequark>
3D printing still requires a CAM, but apparently it's much simpler to do
<wpwrak>
dunno. both have their specialities.
<paul_boddie>
I guess that if you're just doing a layer-by-layer print then the STL file is good enough.
<whitequark>
you need to print supports and such
<paul_boddie>
I saw some 3D printing going on last weekend at a Maker Faire. It was quite the fashion.
<paul_boddie>
A lot of it seemed to involve mostly the same kind of MakerBot device or whatever it is called.
<whitequark>
(fashion) in what sense?
<paul_boddie>
Meaning that a lot of people was doing it. There was some other "maker" stuff going on, too, but 3D printing was all over the place.
<paul_boddie>
s/was/were/
<qi-bot>
paul_boddie meant: "Meaning that a lot of people were doing it. There were some other "maker" stuff going on, too, but 3D printing were all over the place."
<paul_boddie>
Grrr! qi-bot applies global search and replace without the "g" flag!
<paul_boddie>
Not sure if it really cuts it, though.
paul_boddie has quit []
jekhor has quit [Read error: Operation timed out]
<wpwrak>
3d printing is nice. very versatile, quiet, clean. cnc milling on the other hand is: precise, takes many materials, but shapes are limited (no overhangs with a 3 axis mill), noisy, and produces copious amounts of dust
<wpwrak>
so 3d printing is a bit more on the art side while cnc is more on the engineering side
<wpwrak>
now divide potential art critics by the number of potential engineers and you have the ratio :)
arielenter has joined #qi-hardware
<roh>
3d printing is great for prototyping, but you should not expect proper measurements with devices below a few thousand euros
<roh>
fdm isnt great at keeping tolerances down
<roh>
the powder and 'soup' style stuff is better on that.. but hey.. it has its usecases